| Document
image processing is often confused with image processing for desktop
publishing and graphic design. The tools and methods used in each
environment can vary greatly. One can consume many minutes, even
an hour with mouse in hand, preparing a single image for a newsletter,
an ad or an article. In contrast, document image processing is
almost always automated, is performed on batches of images, and
the total processing time for each image is measured in seconds.
In the past, automated image processing required a software
development effort to build a customized facility. Once developed,
the payoff was only a matter of time. Today, the most advanced
image processing products can be configured to perform almost
any image processing task without the need for any programming.
These latest advancements have made sophisticated, highly automated
image processing an affordable option for medium and small companies.
Lets look at a few examples where automated processing of document
images is required.
Privacy
Act Conformance -
Blanking out sections of images. When a form contains personal
information that should not be seen by certain employees, the
original image can be archived and the on-line images are blanked
out in the areas that contain the private information. The process
of blanking out sections is performed automatically on
batches.
Signature
Verification -
Clipping the signature area from a document. Before document
imaging, Banks required their customers to fill out several
signature cards for quick reference by bank personnel. Today,
the checking account application form can be scanned and the
signature section electronically clipped out and stored in a
fast access database for signature verification by tellers in
all branches.
|
Litigation
Evidence Documents - Endorsing
Documents scanned for
litigation must be legible and each page must be electronically
numbered. While despeckling is occasionally used to enhance
image clarity, it can remove lightly printed text, so illegible
documents are normally rescanned with adjusted scanner controls
(e.g. brightness). Automatic document image endorsing is used
to identify each page. These numbers are recorded by the court
for reference and control. A single trial may require tens of
thousands of images as evidence.
Forms
Identification
In some applications, differing forms can be mixed. The forms
can be electronically identified and sorted using image processing
techniques. Once identified, each form image can be further
processed.
The above examples illustrate fully automated applications
of document image processing. Sometimes, image processing requires
the human touch. To maximize throughput while retaining control
on quality, the best semi-automatic systems incorporate large
displays with image enhancing technology (e.g. scale-to-gray)
that display several high-resolution images at a time. Rapid
retrieval rates are also essential.
Tips
for buying a production document image processing system.
If you're shopping for such a system, look for the following
features:
- User-customizable image processing functions without programming
- User-definable single key operation
- Highly functional non-mouse operation.
- On-screen image processing with the ability to abort changes.
- A versatile export facility for interfacing to image management
software
- Ability to combine multiple image processing functions into
a single command without programming (e.g. clip and rescale)
|
